U.S. patent application number 13/548553 was filed with the patent office on 2013-01-17 for adiabatic mode-profile conversion by selective oxidation for photonic integrated circuit.
This patent application is currently assigned to INNOLUME GMBH. The applicant listed for this patent is Alexey Gubenko, Alexey Kovsh, Igor Krestnikov, Daniil Livshits, Sergey Mikhrin, Greg Wojcik. Invention is credited to Alexey Gubenko, Alexey Kovsh, Igor Krestnikov, Daniil Livshits, Sergey Mikhrin, Greg Wojcik.
Application Number | 20130016942 13/548553 |
Document ID | / |
Family ID | 46584368 |
Filed Date | 2013-01-17 |
United States Patent
Application |
20130016942 |
Kind Code |
A1 |
Gubenko; Alexey ; et
al. |
January 17, 2013 |
Adiabatic Mode-Profile Conversion by Selective Oxidation for
Photonic Integrated Circuit
Abstract
Waveguide designs and fabrication methods provide adiabatic
waveguide eigen mode conversion and can be applied to monolithic
vertical integration of active and passive elements in PICs. An
advantage of the designs and methods is a simple fabrication
procedure with only a single etching step in combination with
subsequent well-controllable selective oxidation. As a result,
improved manufacturability and reliability can be achieved.
Inventors: |
Gubenko; Alexey; (Dortmund,
DE) ; Krestnikov; Igor; (Dortmund, DE) ;
Mikhrin; Sergey; (Dortmund, DE) ; Livshits;
Daniil; (Dortmund, DE) ; Wojcik; Greg; (Santa
Clara, CA) ; Kovsh; Alexey; (Munich, DE) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Gubenko; Alexey
Krestnikov; Igor
Mikhrin; Sergey
Livshits; Daniil
Wojcik; Greg
Kovsh; Alexey |
Dortmund
Dortmund
Dortmund
Dortmund
Santa Clara
Munich |
CA |
DE
DE
DE
DE
US
DE |
|
|
Assignee: |
INNOLUME GMBH
Dortmund
DE
|
Family ID: |
46584368 |
Appl. No.: |
13/548553 |
Filed: |
July 13, 2012 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
61507233 |
Jul 13, 2011 |
|
|
|
Current U.S.
Class: |
385/14 ;
257/E21.09; 438/31; 977/774; 977/932 |
Current CPC
Class: |
G02B 6/136 20130101;
H01S 5/1064 20130101; G02B 2006/12078 20130101; G02B 6/1228
20130101; G02B 2006/12097 20130101; H01S 5/2215 20130101 |
Class at
Publication: |
385/14 ; 438/31;
977/774; 977/932; 257/E21.09 |
International
Class: |
G02B 6/12 20060101
G02B006/12; H01L 21/20 20060101 H01L021/20 |
Claims
1. A tapered ridge waveguide formed in a semiconductor layered
structure providing an adiabatic mode-profile conversion by a
lateral oxidation of Al-rich layers, wherein: a) the semiconductor
layered structure is grown on a substrate, and comprises: i) an
active region comprising layers selected from the group consisting
of: a plurality of quantum wells, a plurality of quantum dot
layers, a plurality of quantum dots in a well, and any combination
of a plurality of quantum wells, a plurality of quantum dot layers
and a plurality of quantum dots in a well, for light generation or
to control a spectrum, power, or phase of propagating light by
injected carriers, temperature or an electrical field; ii) a
passive region optimized for low-loss light wave propagation; iii)
a mode-control region, including at least one Al-rich layer,
wherein a refractive index of the mode-control region can be
changed by oxidation, which enables control of an overlapping of an
eigen mode with the active region and the passive region; and iv)
at least one cladding region, the cladding region having refractive
indexes less than a refractive index of the active region and a
refractive index of the passive region; wherein a material and a
thickness of the active region and the passive region are designed
to provide mode localization either in the active region or in the
passive region; and b) the tapered ridge waveguide comprises a
longitudinal single-step ridge waveguide comprising a plurality of
the layers in the semiconductor layered structure and: i) a narrow
section, wherein a width of the narrow section is selected such
that oxidation of the mode-control region results in the
confinement of the eigen mode in the passive region inside the
narrow section; ii) a wide section, wherein a width of the wide
section is selected such that an effective refractive index of the
wide section is negligibly influenced by oxidation of the
mode-control region and therefore the eigen mode is confined in the
active region inside the wide section; and iii) a laterally tapered
section connecting the narrow section and the wide section, wherein
a change of a width of the laterally tapered section provides
gradual optical mode power transfer between the active region and
the passive region.
2. The waveguide of claim 1, wherein the ridge is etched at least
down to a bottom cladding layer to minimize optical losses at a
bending of the waveguide.
3. The waveguide of claim 1, wherein the mode-control region
comprises a step-grading Al-rich multilayer structure.
4. The waveguide of claim 1, wherein the mode-control region is
located within the active region such that complete oxidation of
the mode-control region results in displacement of the eigen mode
from the active region to the passive region.
5. The waveguide of claim 4, wherein the passive region is located
below the active region and the cladding region comprises at least
one bottom cladding layer located between the passive region and
the substrate and at least one top cladding layer located above the
active region.
6. The waveguide of claim 5, further comprising a cap layer located
above the top cladding layer.
7. The waveguide of claim 1, wherein the mode-control region
comprises at least one layer placed below the active region and at
least one layer placed above the active region, such that complete
oxidation of the mode-control region result in displacement of the
eigen mode from the active region to the passive region.
8. The waveguide of claim 7, wherein the passive region is located
below the mode-control region and the cladding region comprises at
least one bottom cladding layer located between the passive region
and the substrate and at least one top cladding layer located above
the mode-control region.
9. The waveguide of claim 8, further comprising a cap layer located
above the top cladding layer.
10. The waveguide of claim 1, wherein the waveguide parameters are
optimized for keeping only a fundamental waveguide mode.
11. The waveguide of claim 1, wherein the laterally tapered section
has a profile selected from the group consisting of: a) a linear
profile wherein a length of the taper is large enough for adiabatic
transfer of the eigen mode between the active region and the
passive region; b) an exponential profile wherein the taper has a
smaller mode transformation loss than with the linear profile and
provides the adiabatic transfer of the eigen mode between the
active region and the passive region at a smaller length of the
taper than in the linear profile; c) a non-exponential curved
profile, wherein the taper has a smaller mode transformation loss
than with the linear profile and provides the adiabatic transfer of
the eigen mode between the active region and the passive region at
a smaller length of the taper than in the linear profile; and d) a
multi-section profile comprising any combinations of a), b, and
c).
12. The waveguide of claim 1, further comprising a doping profile
that enables a realization of electrical conductivity,
p-n-junction(s) and highly doped contact layers.
13. The waveguide of claim 12, wherein the passive region is doped
by an n-type material to reduce power losses.
14. The waveguide of claim 1, wherein a difference between an
effective refractive index for transverse electric polarization and
an effective refractive index for transverse magnetic polarization
is less than 10.sup.-3.
15. The waveguide of claim 1, wherein the cladding region further
comprises a stop-etching layer introduced into a bottom cladding
layer.
16. The waveguide of claim 1, wherein a thickness of the passive
region is thicker than a thickness of the active region.
17. The waveguide of claim 1, wherein the width of the wide section
ranges from approximately 0.5 .mu.m to approximately 5 .mu.m and
the width of the narrow section ranges from approximately 0.3 .mu.m
to approximately 3 .mu.m.
18. The waveguide of claim 1, wherein the at least one Al-rich
layer comprises at least 80% aluminium.
19. A method of fabricating a tapered ridge waveguide for adiabatic
low-loss mode-profile conversion based on oxidation technology,
wherein the fabrication process comprises the following steps: a)
epitaxially growing a layered structure on a substrate, wherein the
structure includes the following regions, wherein each region
represents at least one layer with a certain thickness and
composition: i) an active region comprising a plurality of layers
selected from the group consisting of: a plurality of quantum
wells, a plurality of quantum dot layers, a plurality of quantum
dots in a well, and any combination of a plurality of quantum
wells, a plurality of quantum dot layers and a plurality of quantum
dots in a well, for light generation or to control a spectrum,
power, or phase of propagating light by injected carriers,
temperature or an electrical field; ii) a passive region optimized
for low-loss light wave propagation; iii) a mode-control region
including at least one Al-rich layer, wherein an Al composition of
the Al-rich layer is sufficiently high to be transformed to
(AlGa).sub.xO.sub.y by oxidation; and v) at least one cladding
region, the cladding region having refractive indexes less than a
refractive index of the active region and a refractive index of the
passive region; wherein the composition and thickness of the layers
composing the active, the passive, and the mode control regions,
are designed to provide mode localization either in the active
region if the mode control region is not oxidixed or in the passive
region if the mode control region is oxidixed; b) forming a ridge
waveguide in a single etching step, wherein the waveguide comprises
the following sections along the ridge: a narrow section, a wide
section, and a laterally tapered section, wherein the laterally
tapered section connects the narrow section and the wide section;
and c) selectively oxidizing the mode control region, wherein an
oxidation time and an oxidation temperature are selected such that
an oxidation depth changes an effective refractive index of the
mode control region in the narrow section in order to provide mode
localization in the passive region inside the narrow section;
wherein a thickness of the wide section is selected such that
oxidation of the mode-control region in the wide section weakly
influences effective refractive index of the wide section and the
eigen mode is confined in the active region inside the wide
section; and wherein a change in the width of the laterally tapered
section provides gradual optical mode power transfer between the
active region and the passive region, and wherein the power losses
during the mode transfer can be controlled by the geometrical
parameters of the laterally tapered section.
20. The method of claim 19, wherein the width of the wide section
ranges from approximately 0.5 .mu.m to approximately 5 .mu.m and
the width of the narrow section ranges from approximately 0.3 .mu.m
to approximately 3 .mu.m.
21. The method of claim 19, wherein the at least one Al-rich layer
comprises at least 80% aluminum.
Description
REFERENCE TO RELATED APPLICATIONS
[0001] This application claims one or more inventions which were
disclosed in Provisional Application No. 61/507,233, filed Jul. 13,
2011, entitled "Adiabatic Mode-Profile Conversion by Selective
Oxidation For Photonic Integrated Circuit". The benefit under 35
USC .sctn.119(e) of the United States provisional application is
hereby claimed, and the aforementioned application is hereby
incorporated herein by reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] This invention relates to the field of integrated optics,
and more particularly to methods of monolithic integration of
active devices with passive components.
[0004] 2. Description of Related Art
[0005] Presently many optoelectronic systems are assembled from
separate components that are individually packaged into fiber
modules. These components include, but are not limited to, LEDs,
lasers, amplifiers, modulators, detectors, power splitters,
switchers, filters, and multiplexers. However, the cost of the
components is high mainly because of the package itself, where
coupling optics, temperature stabilization, and precise adjustment
are all required. Moreover, systems based on the discrete
components are power consumable and it is difficult to make them
compact in size. Joining the components into a single-package
configuration, also known as a photonic integration circuit (PIC),
eliminates these disadvantages.
[0006] Photonic integration circuits can be based either on hybrid
or on monolithic integration. Hybrid photonic integrated circuits
bring together optical devices based on different material systems,
for example, an III-V evanescent laser bonded on Si (A. W. Fang et
al., "Electrically pumped hybrid AlGaInAs-silicon evanescent
laser", Optical Express, vol. 14, 9203-9210, 2006). An advantage of
hybrid integration is that each component is optimized for one
specific function, enabling deployment of state of the art
components. However, there are also disadvantages including, but
not limited to, an inefficient light coupling between the
components, different lattice and thermal expansion constants, and
diffusion of impurities between the components.
[0007] On the other hand, monolithic integration joins the devices
based on the same material system, avoids aligning and bonding
problems, and provides exceptional thermal and mechanical
characteristics (see for example U.S. Pat. No. 7,282,311 by Little,
issued Oct. 16, 2006). Taking into account these benefits,
monolithic integration can be preferable for certain applications
with modest integration levels.
[0008] Low-loss optical waveguides are normally needed in PICs for
interconnection and also for some passive components, e.g. spectral
and spatial filters, splitters, delay lines, and chromatic
dispersion compensators. There are a few approaches for monolithic
integration of the passive waveguides including different regrowth
technologies, quantum well intermixing, and vertical twin-waveguide
structure growth.
[0009] The most straightforward passive waveguide integration
technique is epitaxial growth of a second waveguide with the
desired properties after the removal of the original waveguide,
also known as the butt jointregrowth method (see U.S. Pat. No.
4,820,655 by Noda, issued Apr. 11, 1989). The main advantage of
this integration scheme is a high degree of flexibility in the
design, for example, compositions, thicknesses, and doping
concentrations. However, the epitaxial crystal growth at the
abutting locations creates the problem of layer misalignment and
imperfect interfaces (quality and shape) between the active and
passive components, which results in scattering loss, parasitic
optical feedback, and low coupling efficiency. Another regrowth
approach, selective area growth, uses a dielectric mask to inhibit
epi-layer growth during metal organic vapor phase epitaxy (MOVPE)
or metal organic chemical vapor deposition (MOCVD) and, as a
result, to tailor the waveguide properties along its length (see
U.S. Pat. No. 5,543,353 by Suzuki, issued Aug. 6, 1996 and U.S.
Pat. No. 7,060,615 by Glew, issued Jun. 13, 2006). However, the
waveguide properties cannot be strongly changed on a short distance
resulting in additional absorption losses and chirp in the region
of the band edge transition. Moreover, a very precise control of
growth parameters is necessary.
[0010] Another passive waveguide integration method is based on
disordering of quantum wells, also known as quantum well
intermixing (QWI), to locally change band-edges (see U.S. Pat. No.
6,989,286 by Hamilton, issued Jan. 24, 2004). Since the QWI process
only slightly modifies the composition profile and does not change
the average composition, there is a negligible refractive index
discontinuity at the interface between adjacent sections. Different
modifications of the QWI technique, such as impurity free vacancy
disordering (IFVD), impurity induced disordering (IID) and
laser-induced disordering (LID), suffer from their specific
drawbacks, including free-carrier absorption, parasitic
conductivity, residual damage from implantation, inferior quality
of recrystallized material after laser melting, and degradation of
the top surface caused by high-temperature annealing. Taking into
account complexity, irreproducibility, and the poor area
selectivity of the intermixing process, QWI technology is not a
practical method for monolithic integration of multi-functional
optoelectronic devices in PIC (J. H. Marsh, "Quantum well
intermixing", Semiconductor Science Technology, vol. 8, pp.
1136-1155, 1993).
[0011] Vertical twin waveguide structure represents a promising
integration platform technology. This integration technique can be
realized by using either the waveguide modes beating concept or an
adiabatic mode transformation concept. In the first case, the power
transfer results from the bimodal interference between two
supermodes of the vertical twin-waveguide (TG) structure (Y.
Suematsu et al., "Integrated twin-guide AlGaAs laser with
multiheterostructure", IEEE Journal of Quantum Electronics, vol.
11, pp. 457-460, 1975; see also U.S. Pat. No. 5,859,866 by Forrest,
issued Jan. 12, 1999). Despite the fact that active and passive
functions are separated into different vertically displaced
waveguides, all integrated components cannot be well optimized
separately due to a requirement of resonant coupling of both
waveguides. Moreover, performance characteristics of the devices
based on the TG structures are not stable due to mode interaction
and fluctuation in the device structure itself (layer thickness,
composition, dry etching profiles). On the contrary, the adiabatic
mode transformation concept, based on an asymmetric twin-waveguide
(ATG) with tapered couplers, is unaffected by modal interference
(see U.S. Pat. No. 6,282,345 by Schimpe, issued Aug. 28, 2001). The
waveguide is designed in such a way that only one mode exists. To
reduce coupling losses during the power transfer process, the
lateral tapering of the active waveguide at a junction of the
active-passive waveguides is used (see U.S. Pat. No. 5,078,516 by
Kapon, issued Jan. 7, 1992). As the active waveguide rib is
narrowed, the mode profile is smoothly transformed without any loss
of power and, finally, the mode is adiabatically pushed down into
the passive waveguide. This allows the independent optimization of
the active/passive devices in a single epitaxial growth step.
However, there are strict requirements for the etching process (at
least two steps), and for the precision of sub-micron lithography
with a complicated alignment procedure. In addition, ridge
waveguides are rather long, and precise control of taper tips is
required.
[0012] Each of the above-mentioned coupling techniques suffers from
one or more of the following major drawbacks: high optical/coupling
losses, poor manufacturability, high cost, insufficient
reproducibility, and inadequate reliability. Therefore, there is a
need in the art for a novel economical and manufacturable
active-passive coupling technique that permits further progress in
photonic-network communication technology.
SUMMARY OF THE INVENTION
[0013] Waveguide designs and fabrication methods for adiabatic
conversion of waveguide eigen mode provide adiabatic mode-profile
conversion in vertical monolithic integration of active devices
with passive elements into a single photonic integrated circuit. An
advantage of embodiments of the present invention is a simple
fabrication procedure which includes single-step etching in
combination with subsequent well-controllable selective oxidation.
As a result, improved manufacturability and reliability can be
achieved.
[0014] A tapered single-step ridge waveguide, which includes a
multilayer transverse epitaxial structure grown on a substrate,
provides an adiabatic mode-profile conversion by a lateral
oxidation of Al-rich layers.
[0015] The transverse layered structure of the waveguide includes
an active region with a plurality of quantum well, quantum dots in
a well (DWELL) and/or quantum dot layers for creating an optical
gain under current injection, a passive region optimized for
low-loss wave propagation, and a mode-control region, including at
least one Al-rich layer. A refractive index of the mode-control
region can be changed by oxidation, which enables control of an
overlapping of an eigen mode with the active region and the passive
region, and provides anti-degradation protection of other Al-rich
layers. The transverse layered structure also includes at least one
cladding region having refractive indexes less than a refractive
index of the active region and a refractive index of the passive
region. In some preferred embodiments, the refractive index of the
active region is higher than the refractive index of the passive
region. A material and a thickness of the active region and the
passive region are designed to provide mode localization either in
the active region or in the passive region.
[0016] In the longitudinal direction, the single-step ridge
waveguide includes a narrow section, a wide section, and a
laterally tapered section that connects the narrow section and the
wide section. The narrow section has a width that is sufficiently
small such that oxidation of the mode-control region results in the
confinement of the eigen mode in the passive region inside the
narrow section. The wide section has a width sufficiently large
such that an effective refractive index of the wide section is
negligibly influenced by oxidation of the mode-control region and
therefore the eigen mode is confined in the active region inside
the wide section. A change of a width of the lateral taper section
provides gradual optical mode power transfer between the active
region and the passive region, and the power losses during the mode
transfer can be controlled by geometrical parameters of the lateral
taper section.
[0017] A method of fabricating a tapered ridge waveguide includes
the step of epitaxially growing a layered structure on a substrate.
The layered structure includes an active region including a
plurality of quantum well, quantum dot layers, and/or quantum dots
in a well (DWELL) for creating an optical gain under current
injection, a passive region optimized for low-loss wave
propagation, a mode-control region including at least one Al-rich
layer, where an Al composition of this layer is sufficiently high
to be transformed to (AlGa).sub.xO.sub.y by oxidation, and at least
one cladding region, the cladding region having refractive indexes
less than a refractive index of the active region and a refractive
index of the passive region. Each region represents at least one
layer with a certain thickness and composition. The composition and
thickness of the layers composing the active, the passive, and the
mode control regions are designed to provide mode localization
either in the active region, if the mode control region is not
oxidized, or in the passive region, if the mode control region is
oxidized.
[0018] The method also includes the step of forming a ridge
waveguide in a single etching step. The ridge waveguide includes a
narrow section, a wide section, and a laterally tapered section
along the ridge. The laterally tapered section connects the narrow
section and the wide section.
[0019] The method also includes the step of selectively oxidating
the mode control region, where an oxidation time and an oxidation
temperature are selected such that an oxidation depth is large
enough to sufficiently change an effective refractive index of the
mode control region in the narrow section in order to provide mode
localization in the passive region inside the narrow section. The
thickness of the wide section is sufficiently high so that
oxidation of the mode-control region in the wide section weakly
influences effective refractive index of the wide section and the
eigen mode is confined in the active region inside the wide
section. A change in the width of the laterally tapered section
provides gradual optical mode power transfer between the active
region and the passive region, and the power losses during the mode
transfer can be controlled by the geometrical parameters of the
laterally tapered section.
BRIEF DESCRIPTION OF THE DRAWINGS
[0020] FIG. 1a is a schematic view of a transverse structure of a
waveguide of the invention, where the mode-control region is out of
the active region.
[0021] FIG. 1b illustrates the refractive indexes and mode profiles
of the waveguide shown in FIG. 1a before oxidation of the
mode-control region.
[0022] FIG. 1c illustrates the refractive indexes and mode profiles
of the waveguide shown in FIG. 1a after oxidation of the
mode-control region.
[0023] FIG. 2a is a perspective view of a first preferred
embodiment of the present invention. FIG. 2b is a
longitudinal-section of FIG. 2a along the mode-control region.
[0024] FIG. 3a is a vertical cross-section scanning electron
microscopy image along line 3a of FIG. 2a.
[0025] FIG. 3b shows the optical field distribution of the
cross-section shown in FIG. 3a.
[0026] FIG. 3c is a vertical cross-section scanning electron
microscopy image along line 3c of FIG. 2a.
[0027] FIG. 3d shows the optical field distribution of the
cross-section shown in FIG. 3c.
[0028] FIG. 3e is a vertical cross-section scanning electron
microscopy image along line 3e of FIG. 2a.
[0029] FIG. 3f shows the optical field distribution of the
cross-section shown in FIG. 3e.
[0030] FIG. 4a is a plan view of the narrow section of a straight
ridge waveguide.
[0031] FIG. 4b shows the optical field distribution of the
cross-section of the narrow section shown in FIG. 4a.
[0032] FIG. 4c is a plan view of the narrow section of a curved
ridge waveguide.
[0033] FIG. 4d shows the optical field distribution of the
cross-section of the narrow section shown in FIG. 4c.
[0034] FIG. 5a shows the bending losses in dB for a 90.degree. bend
as a function of curvature radius R.sub.b for a curved ridge
waveguide according to FIG. 4c for the different etching depth
D.sub.e at TE-polarization of optical mode.
[0035] FIG. 5b shows the bending losses in dB for a 90.degree. bend
as a function of curvature radius R.sub.b for a curved ridge
waveguide according to FIG. 4c for the TE and TM polarization of
optical mode at fixed etching depth of D.sub.e=3.1 .mu.m.
[0036] FIG. 6 is a perspective view of a second embodiment of the
present invention.
[0037] FIG. 7 is a perspective view of a third embodiment of the
present invention.
DETAILED DESCRIPTION OF THE INVENTION
[0038] In order to overcome the drawbacks of the prior art, the
waveguide design and fabrication method are based on the selective
oxidation technology for adiabatic mode-profile conversion in the
vertical monolithic integration of active devices with passive
elements into a single photonic integrated circuit with improved
manufacturability and reliability. Some optical devices for which
the methods and devices of the present invention could be used
include, but are not limited to, lasers, photodetectors,
modulators, light emitting diodes, amplifiers, detectors, power
splitters, switchers, filters, multiplexers, array waveguide
gratings, and passive waveguides.
[0039] "Adiabatic", as defined herein, means gradually or smoothly,
and with negligible power losses/scattering/interference.
"Adiabatic mode transformation", as defined herein, means gradual
mode transformation with negligible power losses. A "single-step
ridge" and "single-step ridge waveguide", as defined herein, are
ridge structures fabricated in a single etching step/process. This
process forms a structure with upper surfaces at two heights (a
ridge with a single "step"). "Al-rich" layers, as defined herein,
are aluminum containing layers with a high aluminum composition
("rich in Al"). In preferred embodiments, "Al-rich" layers are
layers with an Al composition that is sufficiently high to be
transformed to (AlGa).sub.xO.sub.y by oxidation. In preferred
embodiments, the Al composition in Al-rich layers is greater than
or equal to 80%.
[0040] The design and method address the problem of vertical
monolithic integration of active devices with passive elements. The
devices fulfill three general criteria and provide an effective
active-passive coupling technique promising for use in
monolithically integrated devices such as PIC: [0041] 1) control of
the refractive index profile of the waveguide and, thus, an
overlapping of the eigen mode with the active and passive region by
lateral oxidation of Al-rich layers; [0042] 2) mode localization
either in the active region, or in the passive region depending on
the status of the mode-control region (e.g. oxidized or
non-oxidized mode-control region); and [0043] 3) adiabatic
reversible transfer of the eigen mode between the active region and
the passive region.
[0044] A waveguide structure is depicted schematically in FIG. 1.
The waveguide structure 10, shown in a vertical cross-section in
FIG. 1a, includes an active region 5, a passive region 3, a
mode-control region 4 and surrounding cladding regions 2 and 6. The
mode-control region 4 surrounds the active region 5. In one
preferred embodiment, one or more Al-rich layers act as the
mode-control region 4. High Al composition is preferable for
oxidation (the higher the Al composition the easier (quicker) the
oxidation process). In preferred embodiments, the Al composition of
the Al-rich layers of the mode-control region 4 is between
approximately 80% and 100%. While the mode-control region 4 in this
figure surrounds the active region 5, in other embodiments the
mode-control region 4 may be inserted into the active region (see,
eg., FIG. 7). In other embodiments, the mode-control region 4 is
located only above or below the active region.
[0045] FIG. 1b and FIG. 1c show the refractive indexes 8 and mode
profiles 9 of FIG. 1a in this preferred embodiment, before and
after oxidation of the Al-rich layer in the mode-control region 4,
respectively. In the case of a non-oxidized mode-control region 4,
the optical mode 9 has an effective refractive index 7 higher than
the effective refractive index 8 for the passive region 3 and lower
than the effective refractive index 8 for the active region 5. The
optical mode 9 is thus preferably localized in the active region 5
as illustrated in FIG. 1b. However, the process of wet oxidation
transforms the Al-rich layer into an aluminum oxide
(e.g.--(AlGa).sub.xO.sub.y) and, hence, causes reduction of the
effective refractive index 7 of the mode-control region 4. As a
result, the optical mode 9 has an effective refractive index 7
higher than the effective refractive index for the cladding regions
2 and 6 and lower than the effective refractive index for the
passive region 3. It is consequently mainly localized in the
passive region 3 as shown in FIG. 1c. To achieve such
functionality, the refractive index of the active region 5 should
be higher than the refractive index of the passive region 3, and
the thickness of the passive region 3 should be larger than the
thickness of the active region 5. The waveguide 10 can be designed
to confine only one eigen mode (fundamental mode) in the active
region 5 before lateral oxidation of the mode-control region 4.
After the oxidation of the mode-control region 4 the fundamental
mode will be confined in the passive region 3.
[0046] A preferred embodiment of the present invention is
illustrated in a perspective view in FIG. 2a. As shown in FIG. 2a,
a tapered ridge waveguide 100 grown on a substrate 110 includes a
bottom cladding region 120, a passive region 130, a mode-control
region 140, an active region 150, a top cladding region 160 and a
cap layer 170. The active region 150, which is grown above the
passive region 130, can include a plurality of quantum well,
quantum dots in a well (DWELL), and/or quantum dot layers for
creating a laser, a photodetector, a modulator, etc., while the
passive region 130 is optimized for low-loss wave propagation. In
accordance with FIG. 1a, the cladding regions 120 and 160 have
refractive indexes less than the refractive index of the active
region 150 and the passive region 130. The mode-control region 140
includes two layers, which are preferably Al-rich layers with an Al
composition sufficiently high for oxidation. One of the layers 142
is placed below the active region 150 and the other layer 141 is
placed above the active region 150. The tapered ridge waveguide 100
is designed as a single-mode and single-step ridge waveguide
including a wide section 101 and a narrow section 103 as well as a
laterally tapered section 102 between them. The ridge is etched
through the active region 150 at least down to the bottom cladding
layer 120 to minimize electric device capacity and minimize optical
losses at bending of the waveguide, which is critical for many
devices including, but not limited to, array waveguide gratings and
ring channel filters. To simplify the fabrication process for the
device (one etching step and one oxidation process), the Al-rich
structures of the mode-control region 140 are preferably uniformly
oxidized at a certain length L.sub.ox (oxidation depth), which
results in formation of a Y-branch like oxidation profile as shown
in FIG. 2b. During the oxidation process, Al-rich layers 141 and
142 partly transform into aluminum oxide layers. The oxidation
starts at the perimeter and then goes deeper into the structure.
The longer the oxidation time, the greater the oxidation depth
L.sub.ox. Section 103 is narrow and layers 141 and 142 are fully
oxidized inside this section. Section 101 is wider and layers 141
and 142 are only partly oxidized inside this section (namely to the
depth L.sub.ox from both sides, marked in black). In the middle
part of section 101, layers 141 and 142 are still not oxidized
(marked in white). Note that FIG. 2b is a section of FIG. 2a made
parallel to layer 141 or layer 142. The effective refractive index
of the tapered ridge waveguide is controlled by proper selection of
the waveguide widths D.sub.w (the largest width of wide section
101) and D.sub.n (the width of the narrow section 103) and the
oxidation depth L.sub.ox to provide localization of the eigen mode
90 (see FIGS. 3 and 4) in the passive region 130 for the narrow
section 103 (referred to as the oxidized mode-control region) and
to provide localization of the eigen mode 90 in the active region
150 for the wide section 101 (referred to as the non-oxidized
mode-control region). In other words, the width of the narrow
section is preferably sufficiently small such that oxidation of the
mode-control region results in the confinement of the eigen mode in
the passive region inside the narrow section and a width of the
wide section is sufficiently large such that an effective
refractive index of the wide section is negligibly influenced by
oxidation of the mode-control region and therefore the eigen mode
is confined in the active region inside the wide section. The width
of the wide section should not be too large in order to avoid very
deep oxidation and large electrical capacitance. The width of the
narrow section should not be too small in order to avoid a
considerable overlap of the optical mode with the active region and
bottom cladding layer, which would cause additional internal and
bending loss. In preferred embodiments, the width of the wide
section ranges from approximately 0.5 .mu.m to approximately 5
.mu.m and the width of the narrow section ranges from approximately
0.3 .mu.m to approximately 3 .mu.m. Mode 9 shown in FIG. 1 is the
same mode (in a one-dimensional profile) as mode 90 shown in FIG. 3
(in a two-dimensional profile).
[0047] Vertical cross-sections of the tapered ridge waveguide 100
of FIG. 2 are shown in FIGS. 3a, 3c, and 3e taken along section
surfaces 3a, 3c, and 3e, respectively, as indicated in FIG. 2a.
Corresponding optical field distributions are shown in FIGS. 3b,
3d, and 3f, respectively. Referring to FIG. 3b, the effective
refractive index of the tapered ridge waveguide 100 in the wide
section 101 is negligibly influenced by a finite oxidation depth
L.sub.ox. Therefore, the eigen mode 90 has an effective refractive
index higher than an effective refractive index for the passive
region 130 and lower than an effective refractive index for the
active region 150, and, thus, the optical mode 90 is propagating
primarily in the active region 150.
[0048] When the optical mode 90 starts propagating through the
laterally tapered section 102, the width of the tapered ridge
waveguide 100 is continuously reduced from D.sub.w to D.sub.n,
hence the effective refractive index of this waveguide is
monotonically decreased and the mode profile is smoothly
transformed. As a result, the optical power of the eigen mode is
gradually transferred from the active region 150 into the passive
region 130 as shown in FIG. 3d. When this reduction in the width of
the waveguide of FIG. 2 takes place over a sufficiently long
distance L.sub.t, then the optical mode 90 is adiabatic with
negligible power losses displaced into the passive region 130
(downward in FIG. 3). The longer the distance L.sub.t, the smaller
the losses. According to calculations for L.sub.t=80 .mu.m, the
losses are less than 8%, and for L.sub.t=160 .mu.m, the losses are
less than 1%. In a preferred embodiment of the tapered ridge
waveguide 100, a length of L.sub.t for the laterally tapered
section 102 is more than 350 .mu.m to keep transformation losses
below 0.01%. Finally, according to FIG. 3f, the complete oxidized
mode-control region noticeably reduces the effective refractive
index of the waveguide of FIG. 2 in the narrow section 103. As a
result, the effective refractive index of the optical mode 90
becomes higher than the refractive indexes of the cladding regions
120 and 160 and lower than the refractive index of the passive
region 130 and, as a result, the optical mode 90 mainly propagates
in the passive region 130.
TABLE-US-00001 TABLE 1 Layer structure of a tapered ridge waveguide
with refractive indices according to the first embodiment of FIG.
2. Layer Layer Thick- Refractive Description Composition ness (nm)
index Substrate 110 GaAs -- 3.45 Bottom cladding
Al.sub.0.81Ga.sub.0.19As 2000 3.03 region 120 Passive region 130
Al.sub.0.72Ga.sub.0.28As 1600 3.08 Mode control region:
Al.sub.0.9Ga.sub.0.1As/(AlGa).sub.xO.sub.y 60 2.92/1.6 bottom layer
142 Active region 150 GaAs 150 3.45 Mode control region:
Al.sub.0.9Ga.sub.0.1As/(AlGa).sub.xO.sub.y 60 2.92/1.6 top layer
141 Top cladding Al.sub.0.81Ga.sub.0.19As 1000 3.03 region 160 Cap
layer 170 GaAs 100 3.45
[0049] In one example of a tapered waveguide 100, the epitaxial
wafer of the tapered ridge waveguide 100 is grown in a single
epitaxial process on a substrate 110 of GaAs by molecular beam
epitaxy (MBE) or metal organic chemical vapor deposition (MOCVD).
The compositions of the layers of this waveguide 100 are summarized
in Table 1. In this example of the waveguide shown in FIG. 2, the
active region 150 is a 150 nm-thick GaAs layer, while the low-loss
passive region 130 is a 1600 nm-thick AlGaAs (72%) layer. The top
cladding region 160 of the tapered ridge waveguide is composed of a
1000 nm-thick AlGaAs (81%) layer overgrown with a cap layer 170 of
100 nm-thick GaAs, while the bottom cladding region 120 is a 2000
nm-thick AlGaAs (81%) layer. The mode control region 140 is
composed of two 60 nm-thick AlGaAs (90%) layers. One layer 141 is
placed between interfaces of the top cladding region 160 and the
active region 150 (e.g. above the active region 150), and the other
layer 142 is placed between interfaces of the active region 150 and
the passive region 130 (e.g. below the active region 150). In this
example, the tapered ridge waveguide 100 is designed for an
operating wavelength of 1.3 .mu.m. The single mode tapered ridge
waveguide 100 is fabricated by optical lithography and reactive ion
etching. The device has a stripe width D.sub.w of 3 .mu.m in the
wide section 101 and a stripe width D.sub.n of 1.6 .mu.m in the
narrow section 103. The etching depth, D.sub.e, of the epitaxial
structure is more than 3.1 Note that a stop-etching layer can be
introduced into the bottom cladding layer to provide high accuracy
of the deep etching process. In this example of the waveguide, the
laterally tapered section 102 has a linear profile with a taper
length L.sub.t of 360 .mu.m. A wet lateral selective oxidation
technique is used to fabricate buried dielectric
(AlGa).sub.xO.sub.y layers with a low refractive index. The
oxidation depth L.sub.ox of the Al-rich layers 141 and 142 is
around 0.9 .mu.m. FIGS. 3a, 3c, and 3e also depict the
cross-sectional scanning electron microscopy images of the
fabricated tapered ridge waveguide 100 along section surfaces 3a,
3c, and 3e, respectively. Note that the abundant amount of oxidant
into the (AlGa)O.sub.y layer 141 with the subsequent vertical
oxidation of the AlGaAs (81%) layer 160 results in the taper
oxidation front of the mode control region 140. Regarding
theoretical simulation, such a complex oxidation front can result
only in weak additional confinement of the optical mode and does
not cause any noticeable changes in performance of the tapered
ridge waveguide 100.
[0050] An important component of the modern PIC systems are curved
optical channel waveguides; therefore, the issue of the excess
losses due to bending is actual and important for the present
invention. FIG. 4a shows a straight ridge waveguide 11, where the
field of the optical mode 90 is symmetric about the field peak and
occurs at the center of the passive region 130 of the waveguide 11,
as shown in FIG. 4b. In contrast, in a curved waveguide 12, the
field mode profile is asymmetric, as the optical mode 90 shifts
toward the outward side of the bend curvature (see FIGS. 4c and
4d). This trend gets more pronounced with a smaller bend radius
R.sub.b and the optical mode 90 becomes leaky. Hence, the
continuous radiation of mode power tangentially out of the curved
waveguide 12 as light travels around the bend causes additional
optical losses. Thus, the bending losses dramatically increase with
a decrease in the curvature radius R.sub.b, as shown in FIG. 5a.
Such excess loss can be reduced by increasing the confinement of
the mode field 90. Indeed, if the mode 90 will be weakly confined
in the passive region 130, then the optical mode 90 will tend to
have long exponential tails extending into the cladding region 120,
e.g. the optical mode 90 will suffer from stronger radiation. In
contrast, the increased degree of modal confinement caused by the
deep etching of the waveguide (etching depth at least down to the
bottom cladding region 120, D.sub.e>3 .mu.m) will result in a
decrease in the bending losses (see FIG. 5a). The wider the
waveguide, the higher the effective refractive index, the stronger
confinement of the optical mode, and, therefore, the lower the
bending losses. On the contrary, the thinner the waveguide, the
lower the effective refractive index, the weaker confinement of the
optical mode, and the stronger power scattering on waveguide
sidewalls. Moreover the higher-order modes tend to have more energy
in the exponential tail outside of the passive region 130, causing
larger bending losses, which can be used for effective selection of
the fundamental mode in the case of multimode waveguides.
[0051] Another important aspect is the influence of polarization on
bending losses. Referring to FIG. 5b, proper design optimization of
the waveguide 10 and increasing the confinement of the mode field
enable one to keep a relatively low level of bending losses for
both TM (transverse magnetic) polarization and TE (transverse
electric) polarization even for small bend radiuses R.sub.b<100
.mu.m. In preferred embodiments, an effective refractive index of
the waveguide for TE polarization is close to an effective
refractive index for TM polarization, which can be important for
polarization-independent photonic integrated circuits. In a
preferred embodiment, a difference between an effective refractive
index for transverse electric polarization and an effective
refractive index for transverse magnetic polarization is less than
10.sup.-3. In summary, in preferred embodiments, the waveguide 10
should be designed as a single-mode waveguide with a deep ridge to
minimize optical losses at bending, where negligible polarization
sensitivity is also possible by proper selection of the waveguide
width and the passive region thickness.
[0052] FIG. 6 schematically illustrates a tapered ridge waveguide
200 in another embodiment of the present invention. A difference
between the tapered ridge waveguide 100 and the tapered ridge
waveguide 200 is that, in FIG. 6, the passive region 130 is grown
above the active region 150. According to the general concept, the
ridge waveguide 200 is designed to confine the eigen mode in the
active region 150 in the wide section 101. The continuous reduction
of stripe width of the waveguide of FIG. 6 (laterally tapered
section 102) results in adiabatic displacement of the optical mode
90 from the active region 150 into the passive region 130 (upward
as compared to FIG. 3). Finally, the optical mode is confined in
the passive region 130 in the narrow section 103.
[0053] In another embodiment shown in FIG. 7, a design of a tapered
ridge waveguide 400 is similar to that for the tapered ridge
waveguide 100 except for the mode-control region 140. The
mode-control region 140 is inserted into the active region 150 and
includes at least one Al-rich layer in this embodiment. The tapered
ridge waveguide 400 is designed to localize the eigen mode in the
active region 150 in the wide section 101. The lateral tapering in
the taper section 102 provides adiabatic transfer of optical power
from the active region 150 into the passive region 130 (downward as
compared to FIG. 3). Finally, the optical mode is confined in the
passive region 130 in the narrow section 103.
[0054] Although the laterally tapered section 102 has a linear
profile for all embodiments illustrated herein, this is not
intended to limit the invention to the precise embodiments
disclosed herein.
[0055] The linear taper should be designed to be large enough for
adiabatic transfer of the eigen mode between the active region and
the passive region. Note that tapers of other forms and profiles
may be used within the spirit of the present invention. For
example, a lateral taper with an exponential profile has smaller
mode transformation losses than a linear taper and provides the
adiabatic displacement of the eigen mode between the active region
150 and the passive region 130 at smaller taper lengths L.
Similarly, with a non-exponential curved profile, the taper has a
smaller mode transformation loss and provides the adiabatic
transfer of the eigen mode between the active region and the
passive region at a smaller length of the taper than with a linear
profile. As another example, a two-section taper has a first
section with a linear profile and a second section with an
exponential profile. This two-section taper provides a trade-off
between a linear profile taper and an exponential profile taper.
For example, the first section of this taper results in preliminary
lateral mode confinement, while the second section provides the
adiabatic power transfer between the active region 150 and the
passive region 130 at smaller total taper lengths L.sub.t.
[0056] The devices address the issue of vertical monolithic
integration of active devices with passive elements into a single
photonic integrated circuit, therefore the design rules of doping
in optoelectronic devices is actual. Depending on the exact
application (light emitting diodes, lasers, modulators, passive
waveguides, etc.), various doping profiles of a waveguide structure
of the present invention to realize electrical conductivity,
p-n-junction(s) or highly doped contact layers are possible. For
example, the doping profile of the devices based on the first
embodiment can be a p.sup.+-type doped cap layer (acting as contact
layer) 170, a p-type doped top cladding region 160, an undoped
active region 150, a lightly n-type doped passive region 130, and
an n-type doped bottom cladding region 120 on an n.sup.+-type
substrate 110. Note that a reverse doping profile for a
p.sup.+-type substrate 110 is also possible, however the passive
region 130 should be preferably doped by n-type material to provide
the lowest optical losses during propagation of the optical mode 90
in the passive region 130.
[0057] The so-called wet lateral selective oxidation of Al-rich
layers technology has a unique feature in that it provides the
opportunity to form buried insulating layers with a high structural
quality and with the required electrical and optical parameters.
Moreover, this technique enables smoothing of the sub-micron
surface roughness of the ridge waveguide, which is especially
important for adiabatic low-loss transfer of the optical power. The
present invention also addresses possible solutions to some
critical problems related to the oxidation technique.
[0058] In fact, the accumulated stress and the amount of the
intermediate products generated in the oxidation reaction result in
poor mechanical stability of the oxidized structures. In addition,
the residual hydro-oxides are metastable, which can result in the
undesirable oxidation reaction in the future. Using in-situ
high-temperature annealing allows not only the effectively removal
of the intermediate products, but also the partial conversion of
the amorphous oxide into the more stable polycrystalline phase.
Furthermore, the use of AlGaAs layers with relatively high
Ga-composition also provides improved mechanical stability compared
to pure AlAs layers.
[0059] Another critical issue is reproducibility and uniformity for
oxidation across the epitaxial structure due to the extremely
sensitive compositional, temperature, and doping (level and type)
dependencies of the oxidation rates, especially in the
Al-concentration range of 96-100%. However, the activation energy
for the oxidation reaction of an Al-rich layer demonstrates weak
composition dependency at an Al-composition less than 92% and,
thus, the oxidation rate is insensitive to small deviations of Al.
In combination with diffusion-limited regimes of oxidation, where
the oxidation process is determined by the diffusion of water vapor
through the oxide to the reaction front rather than reaction rate,
the relatively high degree of oxidation selectivity between AlGaAs
layers provides reproducible oxidation. An additional improvement
of oxidation reproducibility can be provided by the short-time
chemical etching before the oxidation process (for example
NH.sub.4OH:H.sub.2O.sub.2:H.sub.2O solution) to remove the surface
damage and contamination caused by the non-chemical etching, for
example of reactive ion etching and dry etching.
[0060] All of the references mentioned herein are hereby
incorporated herein by reference.
[0061] Accordingly, it is to be understood that the embodiments of
the invention herein described are merely illustrative of the
application of the principles of the invention. Reference herein to
details of the illustrated embodiments is not intended to limit the
scope of the claims, which themselves recite those features
regarded as essential to the invention.
* * * * *